Inundation extend mapping for multi-temporal SAR using automatic thresholding and change detection: a case study on Kosi river of India

• Published in: Spatial information research (Online) Vol. 32; no. 3; pp. 311 - 325
• Main Authors: Pinheiro, Greetta, Raj, Aditya, Minz, Sonajharia, Choudhury, Tanupriya, Um, Jung-Sup
• Format: Journal Article
• Language: English
• Published: Singapore Springer Nature Singapore 01.06.2024; 대한공간정보학회
• Subjects: Database Management; Earth and Environmental Science; Geographical Information Systems/Cartography; Geography; Remote Sensing/Photogrammetry; 공학일반; Flood mapping; Automatic thresholding; Change detection; Google Earth Engine; Sentinel-1
• ISSN: 2366-3286; 2366-3294
• DOI: 10.1007/s41324-023-00555-9

The flood occurrence frequency has increased over the years due to climate change, and various state-of-the-art methods have been proposed for flood mapping using Synthetic Aperture Radar (SAR) data. However, whenever there are similarities in the radar backscatter values of permanent water bodies and sand areas, the riverine floods are generally ignored due to high computational complexity. This paper proposes a multi-source data fusion-based model for mapping the Kosi river floodplain areas in the Supaul district of Bihar, India, using both VV and VH bands of Sentinel-1 SAR imagery. The proposed model involves image pre-processing, classification, and post-processing of results to obtain the flood map. The combination of Otsu automatic threshold detection and change detection methods is used for reducing the overestimation of flooded pixels while identifying flood-prone areas. The post-processing involves the identification of high and low-confidence flood regions, riverine floods, generation of flood maps, and estimation of flooded areas. The impact of the flood on the nearby area is captured using multi-temporal images of the Supaul district. The pre-processing, visualizing, processing, and analysis of the results are carried out in Google Earth Engine. The proposed method is suitable for identifying flooding in both non-permanent and permanently low backscattering areas.